Photochemical Pollution Indicators in the Subtropics

Indicators evaluated from the long-term hourly averages of ozone volume fractions at air monitoring sites are proposed as measures of the photochemical pollution exposure at those sites. These indicators are based on the average of the daily maximum-to-minimum ratio during the period of maximum ozone production and are corrected a), for the average maximum to average total ozone ratio (indicator P1) and b), for the number of hours the limit ozone volume ratio limit of 80 ppb was exceeded (indicator P2). These indicators are then combined into P3 as their geometrical mean. The rationale for the introduction of a new set of indicators is based on the suspicion that ozone volume fractions do not provide information either on the total daily ozone that is produced or on the fraction of it that has produced other photochemical pollution components despite that ozone correlates quite well with some of them. Unlike the European stations, where every case in excess of the 80 ppb limit occurs within the April to September "growth period", stations in the subtropics have longer periods (e.g. at Greater Baton Rouge (GBR), USA, for 2001 to 2008) are shifted towards later months June to November (e.g. the Pearl River Delta (PRD) in China for 2006). While GBR and the rural PRD stations exhibit indicators close to those of polluted stations in northern Italy (Po Valley), the (sub)urban PRD stations have high photochemical pollution indicators. However, a surprisingly low indicator level occurs for the coastal Hong Kong stations for reasons possibly attributable to the prevailing easterly winds which bring fresh air and airborne sea-salt particulates. (doi: 10.5562/cca1807)


IMPLICATIONS
The proposed indicators represent a new approach to photochemical pollution problems and they permit a very different view of the air quality at particular sites.The proposed pollution indicators might be of interest to law-and policy-makers in that they give a better representation of pollution-related problems than do simple limiting values.Such indicators should also function well in different climatic zones.

INTRODUCTION
A considerable rise of tropospheric ozone in the subtropics is predicted 1,2 over the next 100 years.Ozone is a strong oxidant that affects living cells and their constituents, especially those involved in photosynthesis and respiration processes, and, hence, vegetation and human health in general.It also affects numerous other materials prone to oxidation.
Ozone, a typical pollutant in the atmospheric boundary layer of regions with substantial sunshine and vehicle traffic, is relatively easy to measure and has been monitored at many sites for long periods of time.We have recently shown that by a systematic comparison of the accumulated and now widely available 1-hour averages of ambient ozone data from sites in Europe, 3 particularly ones in the central Mediterranean region (Italy, Slovenia, Croatia), 4 valuable information on photochemical pollution can be obtained using newly proposed indicators (vide infra).Actually the new indicators are more relevant for photochemical pollution than traditional ozone indices based on daily average or hourly maximum.Since ozone formation is initiated by photo-dissociation of NO 2 and since its further production and destruction is facilitated by the presence of volatile organic compounds (VOCs) and particles, 5  night.The daily maximum hourly or longer-term values are used to define indices, limits, directives and standards for air quality with respect to ozone.However, these entities provide little or no information on the total amount of ozone produced daily nor on the amounts of other photochemical pollution components that are subsequently produced in further reactions of ozone with other pollutants present in the air.Ozone concentrations do correlate quite well with some photochemical indices 6 but these generally represent reactants (precursors) and not the products of reactions with ozone.It is not unusual, for example, that higher altitude stations in remote areas record higher hourly and long-term (8 hours or more) averages than do stations in urban and suburban sites where the significant destruction of ozone decreases the average but often leads to further deterioration of the air quality.Consequently, ozone fractions, on their own, are not valid indicators of photochemical pollution.By assuming that ozone reaction products (e.g.aldehydes, peroxides, radicals and secondary organic aerosols (SOA) 7 ) represent very potent pollution components of the ambient air, new pollution indicators were devised by considering the typical diurnal variation of ozone volume fractions.The indicators are based on calculating the ratios of daily maximum-tominimum hourly average ozone fractions, the maxima being representative of ozone formation and the minima of ozone destruction and, hence, of the formation of potentially hazardous compounds.The average of such daily ratios over time periods of interest (e.g.growth season 1 April -30 September), yields a valid indication of the ozone load along with its reaction products in air.A further correction factor was also applied in order to give a higher weight to the peak ozone values.This was done in two ways: (1), the average ratio is multiplied by the seasonal average of daily maxima divided by the mean for the whole season (indicator P 1 ) or (2) by accounting for the total time a chosen limiting value (e.g., 80 ppb) was exceeded (indicator P 2 ).
One further problem with this procedure had to be solved.When the daily minimum hourly value was measured as zero, thus preventing the calculation of the max/min ratio, the ratio was set to 1.25 times the maximum value.This factor was obtained by considering the detection limit and precision of used ozone monitors.Originally, the minimum value was set at 0.4 instead of 0 ppb, based on the 1 ppb precision of the monitors 4 which proved to be too large and yielded very high ratios.Setting the zero value to 0.8 ppb, which is equivalent to multiplication of the daily maximum by 1.25, was found to be a more appropriate correction factor because frequent zero values in the data were often a result of imprecise measurements at very low volume fractions.
In order to investigate the incidence of high-ozone episodes at latitudes closer to the tropics, we have analyzed the data for the whole year and compared them with those consecutive six months, the "growth season" in mid-latitudes, that would cover most of the high pollution events.Thus by analyzing ambient ozone data for many European stations, 3,4 it was found that the 80 ppb ozone limit had been rarely, if ever, exceeded outside the April-September period.One would expect a different seasonal distribution at lower latitudes.Present investigation is aimed to find out how do the new indicators perform in the subtropics, concerning both their value and their seasonal (and growth-seasonal) dependence.It should be mentioned that in the Fourier transform frequency analysis of long-term ozone data the ratio of yearly to daily frequency intensities correlates nicely with the R-value of investigated monitoring stations. 8,9

EXPERIMENTAL METHODS
The two indicators: P 1 = RM/ A and P 2 = R (1 + 168 t exc / N), where R is the average of daily maximum-to-minimum ratios, M and A the seasonal average daily maximum and average values, t exc the time in hours the limit of 80 ppb was exceeded and N the number of measured hourly average fractions.The scaling factor 168 represents the number of hours per week and was chosen so as to multiply R by two when the limiting value was exceeded for one hour per week on the average.The two indicators have been applied here in the analysis of data from three stations Capitol (CAP), Louisiana State University (LSU) and Port Allen (PAL) in the Greater Baton Rouge (GBR) area 9 in USA for 2001-2008 (http://www.deq.louisiana.gov/portal/Default.aspx?tabid =2420) and fourteen stations in the Pearl River

P P P 
 , introduced recently 4 as more robust to variations than P 1 and P 2 , is also given.

RESULTS
The results for all stations consist of the following annualized items: averages of the ozone volume fraction (A /ppb), the corresponding average of the daily maximum-to-minimum ratios (R ), the average of daily maxima (M /ppb), the time (in hours) the 80 ppb limit is exceeded (t exc /h), the calculated photochemical pollution indicators P 1 and P 2 , their geometric mean and the number of hourly values (N) for a, the whole year and for b, the six consecutive months with the highest photochemical pollution are given for the GBR area in Tables 1a and 1b and for the Pearl River Delta area in Tables 2a and 2b, respectively.
The chosen mid-latitude growth season (1 April to 30 September) covers almost all the events (91 %) within which the 80 ppb limit was exceeded at the three (sub)urban stations in the GBR area and the data are similar to those from polluted sites along the river Po in northern Italy. 4When the period was shifted by one month from 1 May to 31 October no significant change was observed.The reason is that both in April 2003 and October 2005 unusually high ambient ozone fractions were measured.It seems, however, that a longer period such as 7 months covering April to October, would take into account all the pollution episodes in any given year.The data for the 14 PRD-stations (Figure 1) during 2006 reveal three types of sites: first, the rural type stations CH, XI and JI which lie at some distance both from the heavily polluted Guangzhou area as well as from the sea; second, the coastal type stations LI, TM, QU and DC within the Hong Kong area; and third, the (sub)urban stations within affected by the greater metropolitan area of Guangzhou (Table 2).
Ozone data at the rural stations CH, JI and XI show a similar behaviour to the Baton Rouge stations.However, the frequency of high ozone events (above 80 ppb) outside the mid-latitude growth season (April-September) is much higher here (some 30 %) but reduces to 15 % when the six-months period is shifted to June-November.Both the USA and the China data suggest that the period of frequent high ozone events at lower latitudes is shifted to later months compared to the European April-September growth season.In addition, the data indicate that the spread of high ozone events becomes wider the further south one goes: the April-September period covers over 95 % of such events in Europe, 4,5 the same period 91 % in Louisiana, but in the most high ozone abundant June-November period only 84 % of the events in the PRD.It should be stated that June-July is at PRD the typical "wet" season, while October the "dry" and "polluted" season.It is typical of the subtropics that after a long "winter" a short "spring" is followed by a long "summer" characterized by transport of clean and wet ocean air inland.This is more pronounced at lower latitudes (GBR vs. PRD).Three stations LI, TM, QU and DC at the coast its volume fraction (or mass concentration) shows a pronounced diurnal variation with peak values in the early afternoon and minimum values during rush-hours and Croat.Chem.Acta 84 (2011) 11.

Table 1b .
Ozone data for the Baton Rouge area stations for the period 1 April to 30 September with highest photochemical pollution (source: http://www.deq.louisiana.gov/portal/Default.aspx?tabid=2420)

Table 2b .
Equivalent data for the period 1 June to 30 November with highest photochemical pollution

Table 2a .
Ozone data for the Pearl River Delta area (Figure 1, at right) for the whole year 2006 Assessment of photochemical pollution at PRD sites according indicator P 3 for the whole year 2006.